[cups.bugs] CUPS 1.3svn-r6542 and MIT-krb5 => Segfault ?

Rune Andresen andresen-cups at diku.dk
Wed May 30 02:58:36 PDT 2007


Hi.

(I'm new here - sorry if this is the wrong forum)

I'm trying to build/install a Kerberized CUPS, but have a problem.

If I enable gssapi, the scheduler dies with a segfault, when it tries
to create the ccache. The problem appears to be in 'scheduler/ipp.c',
line 7912. Googling the call (krb5_cc_gen_new), returns this:

http://mailman.mit.edu/pipermail/krbdev/2004-March/002570.html

     'Actually I avoided MIT Kerberos's krb5_cc_gen_new because it
      uses mktemp().'

If I change the code to:

    #  ifdef HAVE_HEIMDAL
      if (krb5_cc_gen_new(krb_context, &krb5_fcc_ops, &ccache))
    #  else
      if (krb5_cc_resolve(krb_context, "MEMORY:", &ccache))
    #  endif /* HAVE_HEIMDAL */

then I don't get a segfault. The segfault is regardless of wether I
use 'Negotiate' or 'BasicDigest' as AuthType - as long as gssapi is
enabled when I build CUPS, the scheduler dies when I try to print.

As I'm new to Kerberos/CUPS, I can not eliminate the possibility that
the segfault is due to some mistakes in the configuration - but I
don't think the scheduler should (be able to) die with a segfault when
it hits a bad configuration.

My configuration/system:

Gentoo Linux (current)
MIT-krb5-1.5.2-r1
CUPS 1.3svn-r6542
Gcc 4.1.2

If you need more info or if you can give a hint to what I've made
wrong, please let me know.

Regards,
Rune Andresen





More information about the cups-devel mailing list